Tight Glycemic Control in Critically Ill Children

Abstract

Background: In multicenter studies, tight glycemic control targeting a normal blood glucose level has not been shown to improve outcomes in critically ill adults or children after cardiac surgery. Studies involving critically ill children who have not undergone cardiac surgery are lacking.

Methods: In a 35-center trial, we randomly assigned critically ill children with confirmed hyperglycemia (excluding patients who had undergone cardiac surgery) to one of two ranges of glycemic control: 80 to 110 mg per deciliter (4.4 to 6.1 mmol per liter; lower-target group) or 150 to 180 mg per deciliter (8.3 to 10.0 mmol per liter; higher-target group). Clinicians were guided by continuous glucose monitoring and explicit methods for insulin adjustment. The primary outcome was the number of intensive care unit (ICU)-free days to day 28.

Results: The trial was stopped early, on the recommendation of the data and safety monitoring board, owing to a low likelihood of benefit and evidence of the possibility of harm. Of 713 patients, 360 were randomly assigned to the lower-target group and 353 to the higher-target group. In the intention-to-treat analysis, the median number of ICU-free days did not differ significantly between the lower-target group and the higher-target group (19.4 days [interquartile range {IQR}, 0 to 24.2] and 19.4 days [IQR, 6.7 to 23.9], respectively; P=0.58). In per-protocol analyses, the median time-weighted average glucose level was significantly lower in the lower-target group (109 mg per deciliter [IQR, 102 to 118]; 6.1 mmol per liter [IQR, 5.7 to 6.6]) than in the higher-target group (123 mg per deciliter [IQR, 108 to 142]; 6.8 mmol per liter [IQR, 6.0 to 7.9]; P<0.001). Patients in the lower-target group also had higher rates of health care-associated infections than those in the higher-target group (12 of 349 patients [3.4%] vs. 4 of 349 [1.1%], P=0.04), as well as higher rates of severe hypoglycemia, defined as a blood glucose level below 40 mg per deciliter (2.2 mmol per liter) (18 patients [5.2%] vs. 7 [2.0%], P=0.03). No significant differences were observed in mortality, severity of organ dysfunction, or the number of ventilator-free days.

Conclusions: Critically ill children with hyperglycemia did not benefit from tight glycemic control targeted to a blood glucose level of 80 to 110 mg per deciliter, as compared with a level of 150 to 180 mg per deciliter. (Funded by the National Heart, Lung, and Blood Institute and others; HALF-PINT ClinicalTrials.gov number, NCT01565941 .).

Figure 1. Assessment, Randomization, and Follow-up of the Study Patients

The informed-consent rate was 50% (825 of 1662 patients). Only patients with a measured blood glucose level greater than 130 mg per deciliter were assessed for exclusion criteria. Two additional patients underwent randomization and were in the study when it was stopped early; these patients are not included in the analyses according to the stipulation of the data and safety monitoring board. Additional details are provided in Tables S1 and S2 in the Supplementary Appendix.

Figure 2. Glucose, Insulin, and Nutrition Levels, According to Study Group

Daily data are for the first 8 study days (the median duration of stay in the intensive care unit). Panel A shows time-weighted glucose averages obtained from a linear interpolation of the glucose values that were used to administer the tight-glycemic-control protocol. Open bars indicate either the value used to qualify for the study (Qual) or a partial study day (day 0, the day of randomization), and shaded bars indicate full study days (midnight to 11:59 p.m.). To convert values for glucose to millimoles per liter, multiply by 0.05551. Panel B shows the median time-weighted average glucose level over the entire study period. Panel C shows the daily total insulin delivery, Panel D the daily glucose-infusion rates, Panel E the daily percentage of total nutrition given enterally, and Panel F the daily total nutrition. In each panel, the boxes represent the interquartile range, and the horizontal lines the median. P values for the comparison between groups were calculated with the use of Wilcoxon rank-sum tests (without adjustment for multiple comparisons).